INDEX

Explanations

references to winners and winning events in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

Preceding "by," or is a game-winning context

np_acts-logits-general · gemini-2.0-flash

won by

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

TypedDataSet

-0.68

tagHelperRunner

-0.64

ValueStyle

-0.60

transQ

-0.58

ViewFeatures

-0.57

elemField

-0.56

OGND

-0.56

awtextra

-0.55

 հղումներ

-0.53

featureID

-0.52

POSITIVE LOGITS

 deth

0.46

 winner

0.45

 winners

0.44

 palk

0.44

 victory

0.43

 finalists

0.43

 triumphed

0.43

 victorious

0.42

 winning

0.41

 defeated

0.41

Activations Density 0.148%

references to winners and winning events in various contexts

Preceding "by," or is a game-winning context

won by

No Comments

No Known Activations

references to winners and winning events in various contexts

Preceding "by," or is a game-winning context

won by

No Comments

No Known Activations